07-15-2012 07:11 AM
I have found many here disable the WMM part of QoS to achieve better download speeds in many N wireless routers. It's confusing because part of the Wireless N standard states that WMM must be enabled for N to operate. I know when experimenting with an EA3500 that disabling WMM made a huge difference in increasing my download speeds but it also stopped the N wireless connection, reverting back G.
Here are 3 very good articles that discuss WMM, its theory and implementation:
Quoting from part 3:
There is no doubt that the Wi-Fi Alliance has been successful in getting wireless product manufacturers to support WMM in the majority of today's Wi-Fi products. And WMM seems to provide a benefit for wireless VoIP (or Skype) phones that properly handle incoming and outgoing voice streams with WMM-compliant tagging and processing.
But my experiments show that WMM provides no benefit for improving video (or audio) streaming robustness by shifting available bandwidth from low-priority data traffic to higher-priority media streams. The only way that this will change is if media servers—both standalone and NAS-embedded—implement WMM-compliant priority tagging, something that, at least until this article, they weren't even aware that they needed to do. And it also appears that some Wi-Fi product manufacturers, like Cisco, may have some work to do on their WMM implementations so that they properly allocate bandwidth among prioritized streams."
I wonder if the problems with WMM in the Cisco EA3500 may be due to Cisco not implementing the WMM tags as stated in the conclusion:
"...And it also appears that some Wi-Fi product manufacturers, like Cisco, may have some work to do on their WMM implementations so that they properly allocate bandwidth among prioritized streams."
I hope Cisco will publish how it is implementing WMM and address the drastic slowdown when WMM is enabled on some their N routers.
07-15-2012 10:07 PM
I'm sure you all remember how I said disabling WMM had no effect on my WAN-LAN speed when using my E4200v1 and cheap, WMM certified Ralink RTL8188CE, correct?
Well... seeing this idea again I remembered that my Wrt310nv1 (cascaded to my E4200 on the same subnet) gives me about 20 Mbps less speed (WAN-LAN) compared to my E4200 (72 Mbps PHY rate). So I tried disabling it and...
310n with WMM (65 Mbps PHY rate): 19.75 Mbps
310n with WMM (135 Mbps PHY rate): 23.85 Mbps
310 sans WMM ( 135 Mbps PHY rate): 39.16 Mbps
310n sans WMM (65 Mbps PHY rate): 30.08 Mbps
One thing which I will note is that I use the 310n as an AP/ Switch because its routing functions are defective (i.e port forwarding/triggering and UPnP).
P.S: Disabling WMM doesn't disable 8022.11N for my specific adapter. But it will affect other products (e.g Intel adapters and Apple products).
07-16-2012 06:15 AM
in my case WMM in e4200v2 seems not working
with netgear 3800 when ATV2 stream youtube video,i can see download speed of my NAS reducing
but after I replace netgear 3800 to e4200v2,my NAS download speed is not change during ATV2 stream video
I think there is something wrong with WMM in e4200v2
we should not just disabled it,WMM is a basic function of advanced router(I around 200USD for it),we need to push cisco to improve it
07-16-2012 08:30 AM - edited 07-16-2012 08:32 AM
I think the problems lie in 2 places:
1. Some of the streaming media providers are not tagging their streams according the the N standards so routers, assuming they're functioning correctly, have no way to know which stream gets proirity.
2. The router manufacturers have, in some cases, done a poor job implementing the N standard as far as how their equipment prioritizes the flow of various streams.
I would like for Cisco to tell us exactly how they implement the N standards. It seems strange that, using WMM and having a single file download as the only process needing the wireless bandwidth, the data flow gets severely throttled by an EA3500. It's as if the router is stopping the file download and looking for another stream to give access to, even when no other stream exists.
I am sure different manufacturers use different methods and we can see some routers work as advertized. But a least for myself, the 802.11n standard has yet to deliver on its promises when I used an EA3500 and an Intel b/g/n 5300 wireless card..
I hope this discussion continues. I think we can all benefit and have a better understanding of exactly what we are paying for when we buy a router or wireless card.
07-16-2012 03:49 PM
They need to hire a reachable engineer to monitor these forums for legitimate firmware/hardware issues. ISPs at dslreports and their own forums already have a similar feature. At the rate they're turning out defective firmware, I'm honestly surprised that they haven't done so already.
They also need to make it more clear which issues they are aware of (probably none) so that I know I'm not wasting my time complaining to them about firmware issues.
If Cisco were more dedicated, this forum would have far more less posts of duplicate issues. I personally think it would be highly effective if they tested their routers with consumers before releasing them.... I believe they used to do something like this, but I'm not sure if they still do it.